10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 1
GARY H. JEFFERSON Brandeis University ALBERT G. Z. HU National University of Singapore JIAN SU Peking University
The Sources and Sustainability of China’s Economic Growth
IN 1978, AT THE outset of its economic reform, China was the world’s tenth- largest economy, with a GDP of about $150 billion, or less than 6 percent of U.S. GDP at the time. By 2005, however, China’s economy, at $2.2 trillion, had grown to become the fourth largest in the world, behind only the United States at $12.5 trillion, Japan at $4.5 trillion, and Germany at $2.8 trillion. The above figures, which come from the World Bank, evaluate GDP at current exchange rates and do not take account of differences in the pur- chasing power of currencies. When measured instead at purchasing power parity (PPP), China is already the world’s second-largest economy, with almost $9 trillion in output, nearly three quarters that of the United States. It has been suggested that, at current growth rates, China’s GDP stated in PPP terms could exceed that of the United States as early as 2010.1 When China’s GDP converted at current exchange rates does match that of the United States, assuming that China’s population remains four times the U.S. population, Chinese income per capita will then be but one quar- ter that of the United States. By comparison, the purchasing power of the average Chinese resident will substantially exceed one quarter that of the average U.S. resident, perhaps rising to the vicinity of one half.
The authors appreciate their research collaboration with China’s National Bureau of Statis- tics and specifically with its staff members Liu Yaodong and Qian Jinchang, who helped with data formatting and offered guidance on aspects of the research presented in this paper. We also appreciate the excellent research assistance of Sun Xiaole and Zhao Yan, doctoral candidates at the International Business School at Brandeis University. We acknowledge the support pro- vided by the National Science Foundation (award no. SES-0519902) that helped support the research assistance provided to this project. Finally, comments provided by Barry Bosworth and Gustav Ranis during the panel session were most helpful for revising the paper. 1. See, for example, Holz (2006, p. 41).
1 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 2
2 Brookings Papers on Economic Activity, 2:2006
What changes will have to occur within China’s productive sectors for China’s GDP to match and ultimately surpass that of the United States? Today even China’s coastal industry, the country’s most techno- logically advanced region and sector, lags substantially behind the world technology frontier. Meanwhile a well-known feature of China’s rapid economic transformation is the unequal advance, in terms of technolog- ical change and productivity, of different regions and sectors across this large and populous country. The regions and sectors that lag behind China’s coastal industry also exhibit large disparities in productivity among themselves. These large international and internal productivity gaps represent both advantages and disadvantages for China’s ability to sustain high rates of GDP growth. The key advantage is that both the international gap and the internal gaps continue to provide multiple channels through which catch-up can proceed. A well-known disadvantage of the internal gaps is that the accompanying large differences in income threaten social stabil- ity. A further disadvantage of large internal productivity differences, to the extent they prove persistent, is that much of the burden of China’s catch-up with the United States will fall on coastal industry. That is, if productivi- ties in the regions and sectors outside China’s coastal industry remain far below one quarter that of the United States, then coastal industry will have to achieve productivity levels well above one quarter that of the United States. Coastal industry will have to continue as the locomotive pulling the rest of the economy forward. Indeed, if China is to meet its ambitious goal of output parity with the United States, productivity in coastal indus- try may have to closely approach or even exceed U.S. productivity. Yet the history of other successful developing countries suggests that, as it does so, China’s productivity growth is likely to slow substantially, in turn slowing the country’s overall economic growth. A number of questions emerge from this overview and frame the analy- sis in this paper: Within China, how much does China’s coastal industry lag behind the global frontier? How much do China’s other regions and sectors lag behind coastal industry? Is there evidence of catch-up or con- vergence of these regions and sectors with coastal industry? If so, what are the sources of such change? If instead there are growing disparities, what are the causes? To what extent can one expect that, as China’s coastal industry closes in on the global technology frontier, the productivity growth of China’s own technology frontier will slow? 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 3
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 3
We investigate these questions using panels of industry and firm- level data. However, any research agenda that seeks to assess a country’s medium- to long-term economic growth prospects has to take into account that country’s capacity for institutional adaptation, since institutions shape the incentives and prospects for such growth. This is particularly true for China, which remains engaged in two transitions simultaneously: from a centrally planned to a market economy, and from a less to a more devel- oped country. Therefore we also speculate as to what institutional reforms will most directly bear on China’s ability to close its international and internal productivity gaps. These reforms depend on the ability of China’s political system to formulate and enforce the rules that reassign and clar- ify the property rights needed to sustain investment in technology devel- opment and to facilitate the flow of resources to the regions and sectors offering high returns. During the past quarter century of reform, and largely to the surprise of most observers, China’s economic performance has demonstrated con- siderable resilience. In addition to successfully weathering the Asian financial crises of the late 1990s, China has substantially restructured its state enterprise sector and opened itself to the international economy, including by having adopted World Trade Organization (WTO) rules. For two decades now China has sustained an annual average rate of growth of GDP about 6 percentage points higher than that of the United States (about 9 percent versus 3 percent). If China can sustain that growth advantage into the future, then, assuming no change in exchange rates, its GDP unadjusted for PPP will catch up to that of the United States in twenty-five to thirty years. When China’s GDP does catch up to U.S. GDP, that fact will be of more than symbolic importance. Having established an economic system that is as large, if not as efficient, as that of the United States, China’s consumption of natural resources, its participation in the international trading and financial systems, its contribution to global technological advance, and its influence in international relations and conflict man- agement are likely to approach and in some cases exceed those of the United States. The paper is organized as follows. The next section describes the basic model, partly inspired by Edward Denison’s work, that we use to organize our analysis of China’s catch-up prospects. We next examine the magnitude of the relevant productivity gaps, and we focus on the Chinese 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 4
4 Brookings Papers on Economic Activity, 2:2006
economy’s dynamic catch-up processes for reducing both the international and the internal gaps. We then combine our empirical findings to discuss the prospects and challenges for China’s GDP to catch up with that of the United States during the next twenty-five to thirty years. As already sug- gested, any analysis of China’s catch-up prospects over such a horizon must take into account the role of institutions, including both the con- straints they set and the opportunities they offer for shaping the pace at which the relevant productivity gaps are reduced. Finally, we focus on the political economy of China’s economic growth, and we draw various con- clusions from our analysis, including some policy implications.
The Basic Model: Two Productivity Gaps
In his study of the process by which living standards in the major non- U.S. industrial economies narrowed the gap with, and ultimately caught up to, those in the United States, Denison identified several sources of this catch-up, three of which he viewed as key: resource reallocation, scale economies, and movement toward the international technological frontier (table 1).2 In his study of China’s long-run performance, Angus Maddison cites these same three sources of long-run growth: Countries in this situation of relative backwardness and distance from the technological frontier have a capacity for fast growth if they mobilise and allocate physical and human capital effectively, adapt foreign technology to their factor proportions and utilise the opportunities for specialisation which come from integration into the world economy.3 A close examination of Denison’s results suggests the following lessons: —Within the current group of advanced industrial countries (members of the Organization for Economic Cooperation and Development, or OECD), labor productivity in the initially poorer countries grew faster than it did in the richer countries—a necessary condition for catch-up. —Some labor productivity growth originated with capital accumulation (capital deepening), but for the lower-income economies the most impor- tant source of catch-up was growth in total factor productivity (TFP).
2. Denison (1967). 3. Maddison (1998, p. 17). 10269-01_Jefferson-rev.qxd 1/29/0710:56AMPage5
Table 1. Sources of Growth in Labor Productivity during Catch-Up Average growth in GDP Allocation of residual (percentage points) GDP per hour Source of growth per hour worked,(percentage points) Reduced worked, 1950–62b Improved technology Advances 1950 (percent a Residual resource Economies gap with in Country (U.S. = 100)a year) Labor c Capital (MFP) allocation of scale U.S. knowledge Japan 14 6.45 0.77 1.17 4.57 1.07 1.88 1.41 Italy 32 5.36 0.54 0.57 4.29 1.42 1.22 0.88 0.76 Germany 33 5.15 −0.12 0.93 4.43 1.00 1.59 0.83 0.75 Denmark 43 2.56 −0.11 0.77 1.94 0.67 0.64 −0.27 0.75 France 44 4.80 0.37 0.76 3.67 0.95 1.00 0.74 0.76 Norway 48 3.27 0.02 0.85 2.41 0.92 0.57 0.18 0.76 Belgium 50 2.64 0.36 0.28 2.02 0.51 0.51 0.07 0.76 The Netherlands 53 3.65 0.09 0.78 2.79 0.63 0.77 0.43 0.75 United Kingdom 56 1.63 0.10 0.37 1.18 0.12 0.36 0.04 0.75 United States 100 2.15 0.22 0.60 1.36 0.29 0.36 NA 0.75 Average 41.4 3.95 0.22 0.72 3.03 0.81 0.95 0.36 0.76 (excluding U.S.)
Source: Fagerberg (1994). Reprinted with permission. a. Index using 1970 relative prices. b. Data for Japan are for 1953–61. c. Includes changes in employment, hours of work, age-sex composition, and education. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 6
6 Brookings Papers on Economic Activity, 2:2006
—Among the sources of TFP growth, the creation of scale economies and resource reallocation were most important to the catch-up process. Movement toward the international technology frontier was less important for the group as a whole. Although, statistically, the results in table 1 confirm the relative unim- portance of movement toward the international frontier, arguably it is this factor that drives the other two. That is, without the continuous movement of the advanced industrial sector of a developing economy toward the international frontier, the potential gains from internal resource realloca- tion will eventually be exhausted. Furthermore, establishing scale economies depends substantially on acquiring state-of-the-art technologies that embody the potential to scale up. The international technology frontier is indeed synonymous with innovations that exploit scale economies. Drawing on Denison’s analytical perspective, one can think of China’s growth trajectory as being driven by the ongoing reduction of two pro- ductivity gaps. The first is the international productivity gap, which reflects the substantial distance between the international technology frontier and China’s technology frontier, which we define as the productivity of Chi- nese industry or, more specifically, as the productivity of industry in China’s leading coastal areas. The second is the internal productivity disparity between China’s coastal industrial sector and the country’s lagging agricul- tural and services sectors and between coastal industry and the industrial sec- tors of China’s other regions. Of course, the two gaps are not unrelated. Absent an equivalent increase in the productivity of the lagging sectors, as productivity growth in China’s advanced industrial sector reduces the inter- national productivity gap, it simultaneously must increase the internal pro- ductivity gap, creating the potential for growth through internal technology diffusion and factor reallocation. The catch-up of China’s advanced industrial sector toward the world frontier is fundamentally driven by technological advance, which in turn is driven by the integration of China’s industrial economy with the world economy. This integration has been accelerating, spurred by China’s accession to the WTO in 2001, the surge of foreign direct investment (FDI) into China during the past decade, and the rapid intensification of R&D spending, which facilitates the acquisition and diffusion of tech- nology. Rapid movement of China’s industrial economy toward the inter- national frontier has been the driver of China’s sustained rapid GDP 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 7
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 7
growth. Although labor productivity in China’s advanced industrial sector leads that of other regions and other sectors, in 2002 it was still less than one quarter that in the United States. Thus, even if China’s entire labor force and capital stock were to be efficiently reallocated and were per- forming at the current level of the country’s advanced industrial sector, China’s GDP would still be smaller than U.S. GDP. However, productivity differences across China’s regions and sectors have not diminished during the reform period; indeed, ample evidence suggests that they have widened. China’s catch-up thus will require not only the reallocation of labor and capital to the advanced sectors, but also the diffusion of productivity-enhancing technology in the other direction, to the backward sectors. Several institutional reforms will be needed to support the restructuring and upgrading of the backward regions and sectors, including land ownership reform, reductions in imped- iments to labor mobility and interregional trade, banking and corporate governance reform, and laws governing antitrust, bankruptcy, and mergers and acquisitions. (We examine the functions and political economy of these institutional requirements later in the paper.) If levels of produc- tivity across regions and sectors within China do not converge, China’s coastal industry will bear the burden of catch-up, which will make that catch-up more difficult given the tendency for productivity growth in a developing country to slow as its industrial productivity frontier approaches the world productivity frontier.
Measuring the Productivity Gaps: A First Look
We attempt here to assess the magnitude of China’s international and internal productivity gaps. With respect to the latter, we examine in some detail the gaps in labor productivity between industry and agriculture, and between industry and services, both across China and within each of its four major regions. We report findings using both unadjusted employment data from China’s National Bureau of Statistics (NBS) and data that correct for a possible overcounting of employment in agriculture and undercounting in the other sectors. Finally, we extend the analysis beyond labor productivity to capital and total factor productivity. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 8
8 Brookings Papers on Economic Activity, 2:2006
The International Gap Figure 1 illustrates labor productivity differentials for twenty-seven manufacturing industries at the two-digit standard classification level.4 The figure shows productivity gaps between China’s industries and the international frontier industries, defined as the corresponding industry in the United States or Japan, whichever of the two had the higher labor productivity.5 Comparisons are made for 1995 and 2002, for industry in each of China’s four major regions: coastal, northeastern, central, and western.6 Since the Chinese provincial data are based on a subset of the firm-level data from the large and medium-size enterprise (LME) data set of China’s NBS (thus omitting presumably less efficient small firms), we anticipate some upward bias favoring China in these comparisons. On average for the twenty-seven industries, industrial labor productivity in the coastal region in 2002 was just one quarter that of the international frontier. However, this difference represents a substantial gain compared with 1995, when labor productivity at the world frontier was nine times that of the coast. The figure shows that during this seven-year period, within the coastal region, all but four of the twenty-six industries with available data exhibited catch-up.7 One industry that stands out in figure 1 is the food, beverage, and tobacco industry (5), where the rate of catch-up in 2002 seems substan- tially faster than in China’s other industries. A key reason for this dis- parity is the existence of extremely high profits in the tobacco industry: 20.6 percent of total industrial costs in 2002 compared with an overall
4. The industry productivity data for the United States and Japan are from the Groningen Growth and Development Centre, University of Groningen, The Netherlands (www.ggdc.net/ dseries/60-industry.shtml); the industry data for the thirty-one Chinese provinces, autonomous areas, and municipalities are based on firm-level data from the large and medium-size enter- prise data set compiled by China’s NBS. 5. Among the twenty-seven industries, in 2002 the U.S. industry represented the fron- tier in seventeen, and the Japanese industry in the remaining ten. 6. The coastal provinces and autonomous municipalities (hereafter referred to simply as “provinces”) are Beijing, Fujian, Guangdong, Hainan, Jiangsu, Shandong, Shanghai, Tianjin, and Zhejiang; the northeastern provinces are Heilongjiang, Jilin, and Liaoning; the central provinces are Anhui, Guangxi, Hebei, Henan, Hubei, Hunan, Inner Mongolia, Jiangxi, and Shanxi; and the western provinces are Chongqing, Gansu, Guizhou, Ningxia, Qinghai, Shaanxi, Sichuan, Xinjiang, Xizang, and Yunan. 7. For four of the twenty-seven industries—office machinery (19), insulated wire (20), radio and television receivers (24), and airplanes and spacecraft (29)—data are unavailable for at least one of the four regions. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 9
Figure 1. Chinese Productivity Relative to Productivity at the International Frontier by Region and Industry, 1995 and 2002a
Percent of frontier productivity Coastal
100 2002 80 60 40 1995 20
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Northeastern
100 80 60 40 20
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
Central
100 80 60 40 20
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Western
100 80 60 40 20
5 6 7 8 9 101112131415161718192122232425262728293031
Sources: Groningen Growth and Development Centre, 60-Industry Database, September 2006 (www.ggdc.net); NBS. a. Firms reporting negative value added are excluded; blank entries indicate missing data.Industry codes are as follows: 5, food, drink, and tobacco; 6, textiles; 7, clothing; 8, leather and footwear; 9, wood and products of wood and cork; 10, pulp, paper, and paper products; 11, printing and publishing; 12, mineral oil refining, coke, and nuclear fuel; 13, chemicals; 14, rubber and plastics; 15, nonmetallic mineral products; 16, base metals; 17, fabricated metal products; 18, mechanical engineering; 19, office machinery; 20, insulated wire; 21, other electrical machinery and apparatus not elsewhere classified; 22, electronic valves and tubes; 23, telecommunications equipment; 24, radio and television receivers; 25, scientific instruments; 26, other instruments; 27, motor vehicles; 28, building and repairing of ships and boats; 29, aircraft and spacecraft; 30, railroad equipment and transport equipment; 31, furniture, miscellaneous manufacturing, and recycling. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 10
10 Brookings Papers on Economic Activity, 2:2006
Table 2. Industrial Labor Productivity at the International Frontier and in China, 1995 and 2002a Regionb Year Coastal Northeastern Central Western 1995 Ratio of frontier productivity 8.55 15.87 26.32 33.33 to productivity in China Ratio of coastal productivity 1.00 1.86 3.08 3.90 to regional productivity 2002 Ratio of frontier productivity 4.31 8.93 8.00 10.87 to productivity in China Ratio of coastal productivity 1.00 2.07 1.86 2.52 to regional productivity Decline in ratio of frontier 49.6 43.7 69.6 67.4 productivity to Chinese productivity, 1995–2002 (percent)
Sources: Groningen Growth and Development Centre, 60-Industry Database, September 2006 (www.ggdc.net); National Bureau of Statistics, China; authors’ calculations. a. Data exclude food, beverage, and tobacco industry. b. Data are aggregations of firm-level data.
industrial profit rate of 5.6 percent.8 Also, in that year estimated labor pro- ductivity in China’s tobacco industry exceeded that for overall industry by nearly a factor of ten. For these reasons, which are likely to result from the government’s restrictions on entry to the tobacco industry, we omit the food, beverage, and tobacco industry from our calculations in table 2, which focuses on regional differences in China’s manufacturing productivity in relation to the international frontier.
The Internal Gaps China’s internal productivity gap can be described along two dimensions. The first is the gap between the advanced industrial sector and other, more backward sectors, especially the rural agricultural sector, in which much
8. China Statistical Yearboook 2002, table 14-6. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 11
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 11
of China’s labor force resides as underemployed or surplus labor. We also look at productivity gaps between Chinese industry and the Chinese ser- vices sector, as well as gaps across regions within the services sector. The second dimension of the internal productivity gap is that within industry across regions, and in particular between the advanced industrial sector, primarily concentrated in parts of China’s coastal region, and the rela- tively backward industries located in the northeastern, central, and west- ern regions. We examine the magnitudes of both types of gaps. THE AGRICULTURAL-INDUSTRIAL GAP. Table 3 compares average labor productivity (output per worker) in the agricultural sector broadly defined (agriculture, forestry, and fishing) with that in the industrial (including construction) sector.9 The last column in the table reports the ratio of the two productivities. The table reveals, first, that the agricultural-industrial productivity gap is large. In 2005 the average industrial worker produced more than seven times as much as his or her agricultural counterpart. Moreover, the gap has grown. From 6.1 in 1980, the ratio of industrial to agricultural productivity had shrunk by 1990 to 4.3, but thereafter it grew continuously until, at 7.1 in 2005, it surpassed the 1980 level. REGIONAL AND SECTORAL GAPS. The data in table 3 provide a histori- cal perspective on changes in the agricultural-industrial productivity gap, but they do not provide insight into the variety and range of productivity gaps that exist across regions and sectors, including the services sector. To provide this broader picture, table 4 uses official NBS data to com- pute the gaps for the industry, agriculture, and services sectors across the four regions for 1995 and 2004. We address both the magnitude of the gaps and whether, during 1995–2004, they have tended to widen or nar- row. The table uses productivity in China’s coastal industry, which we designate as China’s technology frontier, as the point of reference in these comparisons. We examine first the gaps in 1995. Ratios of productivity in coastal industry to that of industry in the other three regions ranged from 1.78 (western) to 1.32 (northeastern). The largest ratios are those between coastal industry and the agricultural sector, which range from 4.47 (for
9. Of course, for purposes of analyzing patterns of efficient factor allocation, the relevant measure is marginal, not average productivity. We assume that labor’s output elasticities in the different sectors are not vastly different from one another. 10269-01_Jefferson-rev.qxd 1/29/0710:56AMPage12
Table 3. Output per Worker in Chinese Agriculture and Industry Agriculture, forestry, and fishing Industry and construction Output Employment Output Output Employment Output Ratio of industrial (billions of (millions of per worker (billions of (millions of per worker to agricultural Year yuan) workers) (yuan) yuan) workers) (yuan) output per worker 1980 135.9 291.2 466.8 219.2 77.1 2,844.2 6.09 1985 254.2 311.3 816.4 386.7 103.8 3,725.4 4.56 1990 Unadjusted 501.7 389.1 1,289.3 771.7 138.6 5,569.7 4.32 Adjusteda 319 1,572.7 169 4,566.3 2.90 1995 Unadjusted 1,202.0 355.3 3,383.1 2,868.0 156.6 18,319.7 5.42 Adjusted 300 4,006.7 184 15,587.0 3.89 2000 Unadjusted 1,471.6 360.4 4,083.0 4,555.6 162.2 28,088.0 6.88 Adjusted 278 5,293.5 199 22,892.5 4.32 2005 Unadjusted 2,271.8 339.2 6,698.0 8,620.8 180.9 47,649.6 7.11 Adjusted 258 8,805.4 219 39,364.4 4.47
Sources: China Statistical Abstract 2006, pp. 20, 44; Brandt, Zhu, and Hsieh (forthcoming); authors’ calculations. a. Employment is adjusted for possible official overcounting of agricultural workers. This adjustment (computed by Brandt, Zhu, and Hsieh from 1978–2000) is extended to 2005 by assuming that the annual rate of decline in China’s agricultural employment share from 2000 to 2005 was comparable to its rate of decline from 1990 to 2000. 10269-01_Jefferson-rev.qxd 1/29/0710:56AMPage13
Table 4. Ratios of Labor Productivity in Chinese Coastal Industry to Labor Productivity in Other Regions and Sectors, 1995 and 2004 Region Coastal Northeastern Central Western Real Real Real Real productivity productivity productivity productivity growth ratea growth rate growth rate growth rate (percent a (percent a (percent a (percent a Sector 1995 2004 year) 1995 2004 year) 1995 2004 year) 1995 2004 year) Industry Unadjustedb 1.00c 1.00d 12.51 1.32 0.86 17.89 1.64 1.48 13.77 1.78 1.56 14.22 Adjustede 1.00 1.06 1.29 1.39 Agriculturef Unadjusted 4.47 6.90 8.61 3.80 6.94 6.51 7.05 10.50 9.05 10.22 14.04 10.06 Adjustedg 3.21 4.14 2.73 4.16 5.07 6.30 7.35 8.42 Services 1.05 1.50 6.82 1.70 1.98 9.33 2.24 2.96 7.71 2.42 3.67 6.00
Sources: China Statistical Abstract; NBS (1999); Brandt, Zhu, and Hsieh (forthcoming); authors’ calculations. a. In constant yuan, deflated using sector-specific price indices. b. Includes manufacturing, mining, and construction. c. Output is 23,211 current yuan per worker. d. Output is 65,410 current yuan per worker. e. Includes manufacturing, mining, and electricity production and distribution; excludes construction. Excludes Heilongjiang, Xinjiang, and Yunnan provinces. f. Includes forestry and fishing. g. Agricultural employment is adjusted for possible official overcounting as described in the notes to table 3. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 14
14 Brookings Papers on Economic Activity, 2:2006
coastal agriculture) to 10.22 (for western). The ratios between coastal industry and the services sector range from 1.05 (coastal) to 2.42 (west- ern). For 2004, table 4 shows a tendency for the industry gaps to shrink, particularly that between the coastal and the northeastern industrial sec- tors, where industry in the northeastern region appears to have surpassed that of the coast, even as the coastal industry–regional agricultural gaps increased substantially. The gaps between coastal industry and the services sector also tended to increase. Among the eleven pairwise cases, we find a widening of the gap from 1995 to 2004 in all but three. Although some convergence of labor productivity has occurred within industry, the pro- ductivity gap between industry and the agricultural and services sectors generally increased during 1995–2004. MEASUREMENT ISSUES. The disparities between industry and agricul- ture and between services and agriculture may be somewhat overstated if workers who are temporarily migrating to the cities are included in the agricultural totals. Xiaoquan Ding, and Yang Du and Albert Park, argue that the data in the China Statistical Yearbook overstate the number of workers in agriculture.10 According to Ding, “the official statistics on agriculture employment are based on the registered permanent residence system. Although this system impedes rural residents from obtaining urban registration, it cannot prevent rural residents from moving to cities and working in industries.”11 Ding asserts that many migrant workers liv- ing in cities and those working in township and village enterprises are erroneously classified as agriculture workers. Thomas Rawski and Robert Mead estimate that, in the early 1990s, the overcount may have been as high as 100 million, so that 230 million Chinese workers rather than the reported 330 million were actually working in agriculture.12 Loren Brandt, Xiaodong Zhu, and Chang-Tai Hsieh construct an alterna- tive series of sectoral employment figures to take this possible miscounting into account.13 Their agricultural employment data are constructed by taking the NBS estimate, which is already adjusted for employment in rural town- ship and village enterprises, and further correcting for those working in pri- vate firms or self-employed in nonagricultural activities. This correction results in a substantial shift in employment shares: whereas the NBS data
10. Ding (2001); Du and Park (2005). 11. Ding (2001, p. 23). 12. Rawski and Mead (1998). 13. Brandt, Zhu, and Hsieh (forthcoming). 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 15
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 15
Figure 2. Share of Agriculture in Employment in China, Other Selected Asian Countries, and the United States, 1948–2004
Percent
90 China 80
70 South Korea 60
a 50 China, adjusted Japan 40
30 Taiwan 20
10 United States
1950 1960 1970 1980 1990 2000
Sources: Countries’ national statistical offices. a. Series is adjusted for possible overcounting of agricultural workers in China, as described in the notes to table 3.
for 1995 indicate that 52.2 percent of China’s workforce was employed in agriculture in that year, Brandt, Zhu, and Hsieh’s corrected figures set the share at 44.1 percent. The shares of the industrial and services sectors, recorded by the NBS as 23.0 percent and 24.8 percent, respectively, in 1995 rise to 27.1 percent and 28.8 percent under the corrections. The corrections by Brandt, Zhu, and Hsieh end in 2000; we therefore adjust the 2005 NBS figures by assuming that the decline in China’s agricultural employment share from 2000 to 2005 proceeds at more or less the same rate as in Brandt’s adjusted data for 1990–2000, that is, an annual decrease of 1 per- centage point. Thus we assume that China’s agricultural employment share in 2005 was 34 percent, which roughly corresponds to that in South Korea in 1982 (figure 2).14 To complete the series, we reallocate the agricultural
14. In fact, figure 2, once corrected using Brandt, Zhu, and Hsieh’s estimates of China’s employment shares, tracks very closely with South Korea’s decline over 1962–82 from 63 percent to 39 percent. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 16
16 Brookings Papers on Economic Activity, 2:2006
workers who are dropped in 2005 to the industrial and services sectors in the same proportions as the reallocations to these two sectors for 2000. We first use these corrected agricultural and industry employment fig- ures to adjust the labor productivity calculations shown in table 3.15 With the adjusted figures, we find that the ratio of industrial to agricultural labor productivity for 1995 declines from 5.42 to 3.89; for 2005 the decline is from 7.11 to 4.47. Although the adjusted figures still show an increase in the industry-agriculture productivity gap from 1995 to 2000, this increase is substantially less than implied by the original NBS data. We also use these adjusted employment data to correct the pattern of productivity gaps shown in table 4. Because the employment adjustments by Brandt and coauthors for industry and services are nearly identical for 1995 (1.18 and 1.16, respectively), as are our extensions for 2004 (1.24 and 1.20, respectively), and because the absence of provincial and regional cor- rections requires us to assume that the adjustments are uniform over each of the four regions, we do not recompute the ratios for coastal industry to noncoastal industry or for coastal industry to services. We do recompute the ratios for coastal industry to agriculture, which are shown in table 4. Although, as in table 3, the industry-agriculture gaps using the adjusted data grow substantially less than those using the original NBS data, the results do not change our basic conclusion that overall, in relation to coastal industry, the regional and sectoral productivity gaps grew from 1995 to 2004.16 GAPS BETWEEN BACKWARD AND ADVANCED INDUSTRY: A CLOSER LOOK. A second measurement issue relates to the measures of regional industrial labor productivity using the data from the China Statistical Yearbook. These data show three provinces with implausibly high or low levels of labor productivity in 2004: at the high end are Heilongjiang at 95,195 yuan per worker and Xinjiang at 102,551 yuan per worker. At the low end, Beijing’s labor productivity is 68,126 yuan per worker. To check these productivity data, we use another set of NBS data that has been
15. We further assume that the allocation of excess nonagricultural workers to industry and services is done in accord with the same proportional adjustments made to these sectors in 2000. 16. These results are consistent with those in Naughton (forthcoming, figure 9-2) that show an upward drift in China’s Gini coefficient through the 1980s and 1990s, including a sharp increase for the period 1996–2002. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 17
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 17
compiled for the “above-scale” (guimo yishang) enterprises with annual sales in excess of 5 million yuan.17 In addition to focusing on just those enterprises that regularly report to the NBS, a further advantage of these data is that they exclude construction, a component for which the employment classifications are particularly questionable and whose exclusion allows for comparisons using the conventional definition of industry. When we com- pute the provincial comparisons using these data, we find that the labor pro- ductivity measure for Beijing moves into a plausible range, but those for Heilongjiang and Xinjiang remain implausibly high, as does Yunnan’s in addition, exceeding industry’s average labor productivity by 32, 74, and 50 percent, respectively. Closer inspection of these three provinces shows that each is dominated by either petroleum extraction or the tobacco indus- try, both of which enjoy extraordinary profits that account for their high value added per worker.18 Therefore we drop these three provinces and recalculate the regional labor productivities using only the remaining ones. The productivity gaps for regional industry using these adjusted data are shown in table 4. Unlike the broader industry data, these do not show the northeast surpassing the coast. The adjusted data continue to show the central region lagging further behind the coastal and northeastern regions, and the western region behind all the others. These adjustments do not make a substantial difference to the measures of agriculture and services productivity in relation to coastal industry shown in table 4. However, we note that the more narrow definition of industry, excluding construction and firms that do not qualify as “above scale,” results in significantly higher average productivity for coastal industry (98,624 yuan per worker) than for the broad measure of industry (65,410 yuan per worker). This disparity not only underscores the ten- dency of the sectoral productivity gaps to increase when industry is limited to its more formal definition, but also begins to give some indication of the extent of productivity differences within the industrial sector, for example between formal industry and construction and between the “above scale” firms and the smaller industrial enterprises.
17. China Statistical Yearbook 2005, tables 14-2 and 14-5. 18. Specifically, profit per employee in the petroleum extraction industry is 233,633 yuan per employee; for cigarette manufacturing, it is 184,075 per employee, whereas for total (above-scale) industry, the profit rate is 18,597 yuan per employee (China Statisti- cal Yearbook 2005, table 14-3). 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 18
18 Brookings Papers on Economic Activity, 2:2006
Table 2 also reports regional comparisons for the industrial sector. These measures, too, cover a different set of firms from those in table 4. The data in table 2 are based on calculations of labor productivity for our sample of large and medium-size enterprises, which is still more limited than the larger population of above-scale firms.19 Our subset of this data set also covers manufacturing only (excluding mining and electric power generation) and compares levels for 1995 and 2002, the last year for which data are available in the Groningen data set. A further difference with the industry data in table 4 is that the table 2 data are deflated.20 Although these data are thus not directly comparable to those in table 4, they do provide a useful comparison across regions, in particular by omit- ting the mining and petroleum sectors, which inflate the comprehensive industry productivity measures for the northeastern and western regions. Table 2 shows that the 1995 gaps were large, with ratios of coastal pro- ductivity to regional productivity ranging from 3.90 for the western region to 3.08 for the central region and 1.86 for the northeastern region. By 2002, labor productivity in the central and western regions had reduced the gap with the coastal region: the ratios for that year were 1.86 and 2.52, respectively. In contrast, labor productivity in the northeastern region fell behind that of the coast, with the ratio of productivities rising from 1.86 in 1995 to 2.07 in 2002. These manufacturing data that exclude construction, mining, and power generation show persistent gaps between coastal manufacturing and that in the other three regions. Although they show a narrowing of the gaps between the central and western regions and the coast, the gaps remain large. Combining our results in table 2 and table 4, we find that when industry is defined broadly to include petroleum extraction, the northeast is catching up with coastal industry. When indus- try is limited to manufacturing, however, the northeast exhibits limited or no catch-up.
19. In 2005 industry excluding construction accounted for 86.8 percent of value added of total industry (that is, industry including construction, or what NBS calls “secondary indus- try”). Above-scale enterprises accounted for 87.3 percent and LMEs for 63.8 percent of industry output (excluding construction; China Statistical Yearbook 2005, pp. 52, 489, 512). 20. The regional data are deflated by deflating value added for each of the LMEs using a gross output price deflator constructed from current- and constant-price output deflators reported by each firm. At the provincial level, the firm-level data are aggregated using each firm’s share in total provincial value added as the weight. The regional data are a simple average of the data for the included provinces. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 19
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 19
One shortcoming of our productivity comparisons thus far is that they focus exclusively on labor productivity to the exclusion of capital produc- tivity and the broader measure, total factor productivity. To remedy this exclusion, we use our entire NBS large and medium-size enterprise (LME) data set, which includes firm-level data (including manufacturing and min- ing enterprises), to regress the log values of labor productivity, capital pro- ductivity, and TFP on dummy variables for China’s major regions with and without two-digit industry dummies.21 These industry data include mining and petroleum extraction and power generation. One immediately apparent result, shown in table 5, which estimates the regional differences, and table 6, which summarizes these productivity differentials, is that, as in the industry comparisons shown in table 2 and figure 1, in 1995 the coastal region enjoyed a sizable labor productivity advantage over each of the other three regions. The region that diverges most from the coastal region is the northeastern region, followed by the western and central regions. By 2004 all of these disparities had declined substantially. By contrast, we observe no comparable overall narrowing of the capital productivity gap. In regressions without the industry dummies we observe a consistent increase in the capital productivity gap, as the coast substan- tially increases its capital efficiency relative to the other three regions. In these estimates, therefore, we find an overall convergence of labor produc- tivity with a simultaneous divergence in capital productivity. In construct- ing the TFP measures, the larger weight afforded to labor productivity in part accounts for a pattern of overall convergence. The inclusion of the industry dummies substantially alters the results. For labor productivity, including the industry controls magnifies the pro- ductivity disparities in both 1995 and 2004, although their inclusion does not overturn the result of a robust convergence of industrial labor produc- tivity across regions. The industry dummies have the opposite effect on capital productivity, at least in 2004, tending to substantially reduce the productivity gaps between the coast and each of the other three regions, leaving the regional gaps in capital productivity only slightly altered
21. We use a Cobb-Douglas index with weights of 0.64 for labor and 0.36 for capital. These are the values of the output elasticities of labor and capital, respectively, estimated under the restriction of constant returns to scale. 10269-01_Jefferson-rev.qxd 1/29/0710:56AMPage20
Table 5. Estimates of Log Productivity Differences by Regiona Dependent variable Labor productivity Capital productivity Total factor productivity (lnVA/L) (lnVA/K) (lnTFP) Independent variable 1995 2004 1995 2004 1995 2004 Regressions omitting industry dummy variables Constantb 2.412*** 4.001*** −0.838*** −0.301*** 0.885*** 1.979*** (0.017) (0.015) (0.016) (0.016) (0.015) (0.013) Coastal 0.421*** 0.070*** 0.110*** 0.278*** 0.275*** 0.168*** (0.021) (0.017) (0.019) (0.018) (0.018) (0.025) Northeastern −0.267*** −0.069** −0.390*** −0.275*** −0.325*** −0.166*** (0.030) (0.028) (0.027) (0.032) (0.026) (0.025) Western −0.045 −0.012 −0.106*** −0.099*** −0.074*** −0.053** (0.030) (0.024) (0.027) (0.027) (0.026) (0.021) No. of observations 20,653 27,088 20,653 27,088 20,653 27,088 Adjusted R2 0.046 0.002 0.021 0.025 0.039 0.014 Regressions including industry dummy variables Constantb 1.983*** 3.645*** −0.881*** −0.087** 0.637*** 1.891*** (0.067) (0.039) (0.019) (0.044) (0.061) (0.036) Coastal 0.506*** 0.256*** 0.132*** 0.139*** 0.330*** 0.202*** (0.019) (0.017) (0.019) (0.019) (0.018) (0.015) Northeastern −0.253*** −0.065** −0.392*** −0.300*** −0.318*** −0.175*** (0.028) (0.027) (0.027) (0.030) (0.025) (0.025) Western −0.040 −0.040* −0.111*** −0.116*** −0.073*** −0.076*** (0.028) (0.022) (0.027) (0.025) (0.025) (0.021) No. of observations 20,653 27,088 20,653 27,088 20,653 27,088 Adjusted R2 0.174 0.109 0.066 0.124 0.101 0.076
Source: Authors’ regressions using the NBS large and medium-size enterprise data base. Standard errors are in parentheses. Asterisks indicate statistical significance at the *10 percent, **5 percent, or ***1 percent level. a. The omitted dummy variable is the central region. b. The omitted dummy variable is the textile industry in the central region. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 21
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 21
Table 6. Comparisons of Industrial Productivity Estimates by Region Ratio of coastal productivity to productivity in indicated region Without industry With industry dummy variables dummy variables Dependent variable and region 1995 2004 1995 2004
Labor productivity (VA/L) Coastal 1.00 1.00 1.00 1.00 Central 1.52 1.07 1.66 1.29 Northeastern 1.97 1.15 2.13 1.39 Western 1.58 1.08 1.73 1.34 Capital productivity (VA/K) Coastal 1.00 1.00 1.00 1.00 Central 1.12 1.32 1.14 1.15 Northeastern 1.65 1.74 1.69 1.55 Western 1.24 1.47 1.23 1.29 Total factor productivity (TFP) Coastal 1.00 1.00 1.00 1.00 Central 1.32 1.18 1.39 1.22 Northeastern 1.83 1.39 1.90 1.45 Western 1.42 1.24 1.49 1.31
Source: Authors’ calculations using regression results in table 5.
relative to 1995. These industry effects largely reflect the high concentra- tion of extractive industries, including petroleum and natural gas extrac- tion and petroleum refining, in the northeastern and western regions. These capital-intensive industries, which exhibit high labor productivity, also exhibit low capital productivity. This summary of the results reinforces the notion that industrial labor productivity across China’s regions is converging but that coastal indus- try remains some distance ahead of the other regions. Although the results using industry data that include mining and power generation suggest more rapid catch-up than those using manufacturing alone, the inclusion of these capital-intensive industries also follows a pattern in which capital productivity in the three noncoastal regions is falling fur- ther behind that of the coast, thereby slowing but not reversing the catch-up of TFP. In the following two sections we investigate the dynam- ics of productivity catch-up both internationally and within China’s industrial sector. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 22
22 Brookings Papers on Economic Activity, 2:2006
Chinese Productivity and the International Frontier
A question that is central to the pace and timing of China’s GDP catch- up is how productivity growth in Chinese industry responds to the gap between China’s productivity and the international productivity frontier. Because many sources of productivity change, including resource shifts across industries and regions within China, are commingled in the aggre- gate data, we examine the importance of productivity gaps at the industry level. The firm-level data are aggregated to the industry level for each province, distinguishing twenty-seven industries and thirty-one provinces, so that the unit of observation in the regression is the “province-industry- year.” We relate the rate of growth between 1995 and 2002 of these province-industry productivity observations to the gap between produc- tivity in that province and industry and productivity at the international frontier in 1995. We estimate the following basic equation:
⎡ (VA L) − ( VA L) ⎤ =+αα() ()1 ⎣lnij,,2002 ln ij ,, 1995 ⎦ 0 1ln GAP_FRONTij,, 1995 + α ⎡ (VA L) − (VA L) ⎤ + ε 2 ⎣ln FRONT,,j 20002ln FRONT,,j 1995 ⎦ ,
= − where ln(GAP_FRONTi,j,1995) ln(VA/L)FRONT,j,1995, ln(VA/L)i,j,1995, and i indexes provinces and j industries. (The rates of growth are annualized.) To test for regional differences in the response, we include dummy values α of 1 for three of the four regions, where the dummy variables interact with the 1995 productivity gaps. α Our priors are that 1 > 0, reflecting the fact that industries and regions that are further behind the international productivity frontier can make big- ger gains by exploiting the frontier technology, either by imitation or by α importing technology or capital. One might anticipate that the sign on 2 would likewise be positive, indicating that the more rapid is productivity growth during 1995–2002 for a frontier industry, the more generally avail- able the useful technology and spillovers are in its lagging Chinese counter- part industry during the same period. Alternatively, China’s comparative advantage may be greatest in industries such as textiles, apparel, and footwear, where productivity growth in the advanced industrial economies is slow. In this case such Chinese industries might grow rapidly, moderniz- 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 23
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 23 α α ing in the process, leading to a negative 2. Similarly, a negative 2 would also arise where productivity changes were exceptionally rapid in the frontier industries, providing little opportunity for Chinese firms to begin to catch up technologically, discouraging modernization. As these consid- erations suggest, the regression results should be interpreted as casting light only on medium-term responses in China’s recent development. They are informative about the path that China is on but cannot be used with confidence to infer conditions well outside the data, such as long-run equilibrium conditions. The estimation results, shown in table 7, are robust to alternative spec- ifications and samples, showing that the rate of industrial productivity growth during 1995–2002 rises monotonically with the distance of the relevant industry from the corresponding frontier productivity level in 1995. The quadratic term becomes highly statistically significant when the constant, which itself is generally not highly significant, is constrained to equal zero. The findings in table 7 are consistent with Denison’s finding (table 1) that the rate of labor productivity growth in catch-up countries slows as these countries move toward the international productivity frontier. We further find, as shown by the large coefficient on the variable that interacts China’s coastal region dummy with the gap variable, that coastal firms generally enjoy higher rates of productivity growth than do firms in the other three regions for every level of the productivity gap. The results reported for regression 7-5 in table 7 are mapped into figure 3, which shows how productivity growth in both the coastal and other regions relates to an industry’s productivity gap, expressed as the ratio of frontier productivity to Chinese productivity in a given industry, assuming a 2 percent annual growth rate of productivity at the frontier. The figure illustrates the potential importance of pure technological catch-up at the firm level. The effects of gaps are highly significant, and the average pro- ductivity improvement of coastal industries in the face of international gaps is substantial at the level of the gap observed in most industries in 1995. For example, with a ratio between industry productivities of 10, which is smaller than that in many industries in that year, the implied rate of labor productivity growth is 11 percent a year in the coastal region, indicating a rapid reduction of such industry gaps even with substantial growth in frontier productivity. The predicted growth in productivity for a comparable gap in other regions is lower but still substantial (roughly 10269-01_Jefferson-rev.qxd 1/29/0710:56AMPage24
Table 7. Estimates of the Response of Labor Productivity Growth to the International Productivity Gap, 1995–2002a Regression Independent variable 7-1 7-2 7-3 7-4 7-5 Constant −0.077** −0.069* −0.064* (0.037) (0.038) (0.038) ln(GAP_FRONTi,j,1995 ) 0.085*** 0.070*** 0.067*** 0.020** 0.026*** (0.024) (0.025) (0.025) (0.008) (0.009) × ln(GAP_FRONTi,j,1995 ) coastal 0.015*** 0.014*** 0.017*** 0.012*** (0.004) (0.004) (0.004) (0.004) × ln(GAP_FRONTi,j,1995 ) northeastern 0.003 0.004 (0.004) (0.004) × ln(GAP_FRONTi,j,1995 ) western 0.003 0.004 (0.005) (0.004) 2 − − − [ln(GAP_FRONTi,j,1995 )] 0.004 0.002 0.0001 0.006*** 0.005** (0.004) (0.004) (0.004) (0.002) (0.002) − − − − − − [ln(VA/L)FRONT,j,2002 ln(VA/L)FRONT,j,1995 ] 0.159*** 0.196*** 0.189*** 0.204*** 0.196*** (0.052) (0.049) (0.048) (0.050) (0.050) Adjusted R2 0.240 0.259 0.239 0.622 0.621
Source: Authors’ regressions using the NBS large and medium-size enterprise data set. a. The dependent variable is the annual growth rate of labor productivity between 1995 and 2002, in percent a year. All regressions are on 589 observations. China observations for which VA/L > 1,000,000 yuan per worker are omitted. Standard errors are in parentheses. Asterisks indicate statistical significance at the ***1 percent, **5 percent, or *10 percent level. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 25
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 25
Figure 3. Response of the Chinese Productivity Growth Rate to the Gap with the International Frontiera
Growth rate of Chinese productivity (percent a year)
22 20 18 Coastal region 16 14 Other regions 12 10 8 6 4 2
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
Ratio of international frontier productivity to Chinese productivity
Source: Authors’ calculations using regression results in table 7. a. Calculated assuming a 2 percent productivity growth rate at the international frontier, indicated by the horizontal line.
8 percent). The estimation results shown in table 7 and their illustration in figure 3 also imply that, at least for manufacturing, the northeastern, central, and western regions may enjoy rapid productivity growth but will not fully catch up to the coast, at least in the medium term. As the results show, productivity growth in these regions will grow as fast as that in coastal manufacturing only as long as a substantial productivity gap persists. Factors that may explain this persistent disparity between the coastal and other regions include the concentration of FDI and R&D spending in the coastal region and the better development of institutional arrange- ments, including the legal system and human capital development in the coastal region. Together these factors may enable coastal industry to take greater advantage of international technology than industry in other regions can, even though its gap with the world frontier may be consider- ably smaller than those of industry in other regions. We return to these issues later in the paper. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 26
26 Brookings Papers on Economic Activity, 2:2006
Sources of Internal Productivity Growth
Our investigation of the responsiveness of labor productivity growth in China’s domestic industry to international productivity gaps has shown, with the existing large gap, an initial tendency for sustained labor produc- tivity growth and catch-up, particularly in the coastal region. We also find, in tables 2, 4, 5, and 6, evidence within China’s industrial sector of catch- up with coastal industry by the other regions, at least for labor productivity and TFP, if not for capital productivity. This part of the paper investigates the processes through which Chinese firms may or may not respond to pro- ductivity differentials within Chinese industry by closing the internal pro- ductivity gap. The analysis examines the following issues: What are the contributions of labor reallocation and capital accumulation to produc- tivity growth? Can evidence be found of improved allocative efficiency within China’s industrial sector, that is, a closing of productivity gaps arising from the reallocation of labor and investment to firms that offer higher returns? And what is the contribution of the exit and entry of firms to industrial productivity growth?
The Contribution of Labor Reallocation As reported above, we find large differences in labor productivity among sectors and regions within China. Given these differences, the realloca- tion of labor from low- to high-productivity sectors or firms could have substantial effects on aggregate output and productivity. To clarify the potential importance of this mechanism for explaining the rapid growth of Chinese output in the last decade and its potential importance for future growth, we consider a two-sector model in which labor productivity in
agriculture is designated Pa and that in industry gPa. Assuming that neither productivity in agriculture nor productivity in industry changes significantly with the reallocation of labor, moving one unit of labor from agriculture to − industry increases output by (g 1)Pa. β Taking the labor force L0 as given, with an initial fraction of L employed in agriculture, aggregate output is
=+−[]ββ() =+−[] ββ() QPaa11 gPL00 gPL a. 10269-01_Jefferson-rev.qxd 1/29/07 10:56 AM Page 27
Gary H. Jefferson, Albert G. Z. Hu, and Jian Su 27
Figure 4. Impact of Labor Reallocation on GDP Growth for Alternative Shares of Employment in Low-Income Sectorsa
Contribution to GDP growth rate (percentage points)
2.5
2.0 β = 0.8 1.5
1.0 β = 0.5
0.5 β = 0.2
23456789 Ratio of industrial to agricultural productivity
Source: Authors’ calculations. a. Assumes that 1 percent of the low-income labor force moves into the high-income sector each year; β indicates the initial share of the labor force in the low-income sector.
If labor moves out of agriculture at a rate of b percent a year (that is, the percentage rate of change of β is −b), the percentage rate of change of Q is simply